idx_level_0 = pd.date_range('2020-01-01', '2020-04-01', freq = 'M')
idx_level_1 = pd.date_range('2020-04-01', '2020-07-01', freq = 'M')
idx_dates = pd.MultiIndex.from_product([idx_level_0, idx_level_1], names = ['Event_Date', 'Observation_Date'])
ser_info_dated = pd.Series(range(len(idx_level_0) * len(idx_level_1)), index = idx_dates, name = 'Some_Values') / 33
list_levels_dates = sorted(list(set(idx_level_0) | set(idx_level_1)))
dict_to_numbers = dict(zip(list_levels_dates, range(len(list_levels_dates))))
df_info_numbered = ser_info_dated.reset_index().replace({'Event_Date': dict_to_numbers, 'Observation_Date': dict_to_numbers})
df_info_downcasted = df_info_numbered.copy()
df_info_downcasted[['Event_Date', 'Observation_Date']] = df_info_downcasted[['Event_Date', 'Observation_Date']].astype('int16')
这似乎是一个成功:
print('df_info_downcasted column types:\n', df_info_downcasted.dtypes)
结果表明:
df_info_downcasted column types:
Event_Date int16
Observation_Date int16
Some_Values float64
ser_info_downcasted = df_info_downcasted.set_index(['Event_Date', 'Observation_Date']).squeeze()
print('ser_info_downcasted index level 0 type: ', ser_info_downcasted.index.levels[0].dtype)
print('ser_info_downcasted index level 1 type: ', ser_info_downcasted.index.levels[1].dtype)
ser_info_downcasted index level 0 type: int64
ser_info_downcasted index level 1 type: int64
ser_info_astyped = ser_info_downcasted.copy()
ser_info_astyped.index = ser_info_astyped.index.set_levels(ser_info_astyped.index.levels[0].astype('int16'), level = 0)
ser_info_astyped.index = ser_info_astyped.index.set_levels(ser_info_astyped.index.levels[1].astype('int16'), level = 1)
print('ser_info_astyped index level 0 type: ', ser_info_astyped.index.levels[0].dtype)
print('ser_info_astyped index level 1 type: ', ser_info_astyped.index.levels[1].dtype)
ser_info_astyped index level 0 type: int64
ser_info_astyped index level 1 type: int64
目前没有回答
相关问题 更多 >
编程相关推荐